计算机工程与应用 ›› 2015, Vol. 51 ›› Issue (21): 52-57.

• 理论研究、研发设计 • 上一篇    下一篇

模拟电路故障诊断新方法

谢  涛1,2,何怡刚3,李  珩4   

  1. 1.湖南科技大学 计算机科学与工程学院,湖南 湘潭 411201
    2.湖南大学 电气与信息工程学院,长沙 410082
    3.合肥工业大学 电气与自动化工程学院,合肥 230009
    4.中南林业科技大学 外国语学院,长沙 410004
  • 出版日期:2015-11-01 发布日期:2015-11-16

New method of analog circuits fault diagnosis

XIE Tao1,2, HE Yigang3, LI Heng4   

  1. 1.School of Computer Science and Engineering, Hunan University of Science and Technology, Xiangtan, Hunan 411201, China
    2.College of Electrical and Information Engineering, Hunan University, Changsha 410082, China
    3.School of Electrical Engineering and Automation, Hefei University of Technology, Hefei 230009, China
    4.College of Foreign Languages, Central South University of Forestry and Technology, Changsha 410004, China
  • Online:2015-11-01 Published:2015-11-16

摘要: 提出了一种新颖的基于多小波神经网络的模拟电路故障诊断方法。介绍了多小波的原理,分析了多小波神经网络的结构、逼近性质及多小波神经网络的算法,提出了用多小波来处理故障信号,提取故障特征向量输入给神经网络,从而进行模拟电路故障诊断。由于多小波函数具有连续、对称性及支撑集短等一系列优点,所以用多小波神经网络来进行模拟电路故障诊断比一般的小波神经网络具有诊断精度高、诊断速度快的优点。给出了仿真诊断实例,验证了该方法的有效性。

关键词: 模拟电路, 故障诊断, 故障特征向量, 多小波变换, 神经网络

Abstract: This paper proposes a novel method of analog circuits fault diagnosis based on multi-wavelet neural network. It introduces the principle of multi-wavelet, analyzes the structure of multi-wavelet neural network and the approximation properties and the algorithm of multi-wavelet neural network. It uses multi-wavelet to deal with fault signal and extracts fault feature vector input to the neural network, thus diagnoses analog circuits fault. Since the multi-wavelet function is provided with a series of advantages such as continuous and symmetry as well as support set that is short and so on, using multi-wavelet neural network to analog circuits fault diagnosis is provided with advantages such as high accuracy and fast speed of diagnosis than the general wavelet neural network. It gives a simulation diagnosis example, and the effectiveness of this method is verified.

Key words: analog circuits, fault diagnosis, fault feature vector, multi-wavelet transfer, neural network